Does immigration to Thailand reduce the wages of Thai workers?

8/9/2019 Does immigration to Thailand reduce the wages of Thai workers?

1/20

Does immigration to Thailand reduce the wages of Thai workers?

John Bryant and Pungpond Rukumnuaykit

Institute for Population and Social Research

Mahidol UniversityThailand

Data from a Thai campaign to register irregular migrants offer a rare opportunity to study

the labor market effects of immigration in a developing country. We use the registrationdata, plus census and survey data on Thais, to study how immigration has affected wages,

employment, and domestic labor migration in Thailand. Essentially we test whether, all

else equal, Thais living in places with more immigrants have different labor marketoutcomes from other Thais. We allow for endogenous migration, whereby immigrants are

disproportionately attracted to areas with higher wages, by using distance to the

Myanmar border as an instrument for migrant intensity. We allow for geographicalspillovers by estimating our model at two levels of geographical aggregation, and by

constructing a model with spatial lags. We also test whether Thais avoid migrating into

areas that have received more immigrants. Our results suggest that immigration has

reduced the wages of Thais. We find no evidence that immigration reduces employment,

or that it affects internal migration.


2/20

Introduction

Eighty-three percent of Thais believe that immigration to Thailand reduces local wages 1.

Results from empirical studies from developed countries suggest that any reductions arelikely to be small. A meta-analysis of 18 such studies suggests that a one percentage point

rise in the share of migrants in the workforce depresses native wages by only 0.119

percent (Longhi, Nijkamp, and Poot 2005: 472). But it is hazardous to extrapolate fromthese studies, since developing countries differ from developed ones in ways that are

likely to affect the migration-wage relationship. For instance, in Thailand, as in many

developing countries, minimum wages are rarely binding (Falter, No date), and a

substantial proportion of the local workforce does the dirty or dangerous jobs typicallyassociated with migrants.

Evaluating the labor market impacts of immigration to a developing country is typicallyimpossible because of the absence of adequate data on migrants, including irregular

migrants. Thailand is, however, a partial exception. Beginning in the 1990s, Thailand has

received hundreds of thousands of migrants from Cambodia, Laos, and Myanmar, andsmaller numbers from China, Vietnam, and elsewhere. Although these flows include

people who are essentially refugees, immigration to Thailand also reflects wage

differentials: Thailands GDP per capita exceeds GDP per capita in Cambodia, Laos, andMyanmar by about the same multiple that GDP per capita in the United States exceeds

GDP per capita in Mexico. Virtually all migrants from neighboring countries have

entered the country illegally. In 2004, the Thai government allowed Cambodian, Lao, and

Myanmar migrants to register for permits giving them the right to live and work inThailand for one year. Because the terms offered were relatively favorable, the

registration appears to have captured a significant proportion of migrants in Thailand.

The registration data thus offer a rare opportunity to evaluate the effect of immigration on

the wages of natives in a developing country.

We identify the labor market effects of immigration using geographical variation in

migrant intensity. This is the most common strategy used in developed-country studies,

and is the one most suited to our data on migrants, a single cross-section with detailedgeographical information.. Essentially we test whether, all else equal, areas with

unusually high concentrations of migrants have unusually low wages. The test is

complicated by the fact that migrants are attracted to areas with high wages. We addressthis issue by using distance to the border as an instrument for migrant intensity. We also

test whether the effects are strongest for low-skilled Thai workers, and examine other

labor market outcomes such as employment. If native workers are deterred frommigrating into areas that have received large numbers of immigrants, this can transmit the

labor market impacts to other parts of Thailand. We therefore test whether immigration

affects internal migration by Thais. Our results suggest that immigration does lower Thai

wages, but does not lower employment rates or effect internal migration.

1 Unpublished tables from a poll of 4,148 Thais conducted by Assumption University between 25

November and 1 December 2006. The poll was supported by the International Labour Organization and the

United Nations Development Fund for Women.


3/20

Theory and evidence on immigration and wages

In the most basic model of the labor market effects of immigration, the addition ofimmigrants to the labor force shifts the labor supply curve outward, which leads to a

reduction in wages, which induces natives to reduce their labor supply. In the long run,

the labor demand curve also shifts outwards, as firms invest in new capital for theadditional workers. Under constant returns to scale, wages and native labor supply

eventually return to their pre-immigration levels (Altonji and Card 1991). Labor market

effects become more complicated, however, when factors such as the characteristics of

immigrants, native migration patterns, and labor market institutions are incorporated intothe analysis.

The extent to which immigrants actually compete with natives depends on the skills ofimmigrants and natives, and on regulations faced by immigrants. In the United States and

Europe, immigrants span the entire range of education levels (Angrist and Kugler 2003;

Card 2005). In Thailand, almost all immigrants from the main sending countries ofCambodia, Laos, and Myanmar have limited educations. Many migrants to Thailand also

have difficulty speaking the Thai language. In the United States and Europe, the majority

of migrants are legally permitted to reside in the country2, while in Thailand mostmigrants have a much weaker legal position. Migrants who register but had entered the

country illegally are still technically in violation of Thailands Immigration Act, and the

governments long-term stance towards immigrants remains uncertain (Pearson et al

2006: 27-29). Registered migrants are also prohibited from performing skilledoccupations. Moreover, as discussed below, only about a half of all immigrants were

registered in 2004. The weak legal status of migrants presumably deters employers from

investing in their skills, or promoting them. This means migrants most resemble unskilled

Thais, though they differ even from them.

When an area receives immigration, native workers may react by moving out of the area,

or by refraining from moving in. Compensatory migration by natives would reduce the

outward shift in the labor supply curve in the immigrant-receiving areas, but shift laborsupply curves elsewhere in the country. This could explain why studies that use

geographical variation in immigrant shares to identify labor market effects have generally

found small effects. Evidence on the importance of compensatory migration is, however,mixed. Card and DiNardo (2000) find no evidence for the existence of compensatory

migration in United States. Hatton and Tani (2005) find evidence of substantial

compensatory migration in Britain, but only when they restrict their sample to southernEngland. When Borjas (2003) tests for the labor market effects of immigration using

variation across time and age-groups, rather than geographical areas, he finds strong

effects. He attributes the strength of these effects to the fact that his estimates are not

subject to downward biases because of compensatory migration. However, Friedbergs

2Undocumented migrants constitute around 30 percent of all migrants in the United States (Passel 2005: 2),

and around 16 percent in Europe (Mansoor 2007: Table 1.2]


4/20

(2001) study of Israel in the early 1990s is almost as well protected from the biases due tocompensatory migration, since she uses variation in migrant share across occupations, yet

she finds that immigration inflows equivalent to 12 percent of the Israeli population had

no effect on native wages.

Data and methods

Main data sources

Our estimates of migrant numbers are based on data from a registration campaign for

migrants from Cambodia, Laos, and Myanmar in 20043. Altogether, 0.82 millionmigrants obtained work permits, of whom 13 percent were from Cambodia, 13 percent

from Laos, and 74 percent from Myanmar. Only migrants aged 15 and over were

permitted to obtain work permits. Two-thirds of migrants were aged less than 30, and 45percent were women. Many migrants and employers avoided the registration because

they did not wish to identify themselves to the authorities or to spend the necessary time

and money. In some places, the process was also poorly publicized. However, the

registration did provide migrants with some protection from police harassment, andimproved their chances of accessing health care and schooling for themselves and their

dependants (Pearson 2006: 27-29). It also enrolled far more migrants than similar

campaigns before or after (Huguet and Punpuing 2005: 34). Officials and non-

governmental organizations typically assume that the registration campaign coveredaround half the irregular migrants in the country. This is low compared with the

approximately 90 percent coverage of illegal immigrants achieved in the US Population

Census in 2000 (Card and Lewis 2005: 4). But it is considerably better than the 10 -20percent coverage achieved by the Thai Population Census in 20004.Moreover, the

analyses presented in this paper are based on the relative distribution of migrants across

districts, rather than absolute numbers, which helps reduce biases due to under-reporting.

[Table 1 here]

Our main source of data on Thai workers is four rounds of the Labor Force Survey

carried out by the Thai National Statistical Office in 2004. We exclude government

employees5, because these people do not compete with migrants, and their wages are

unlikely to be affected by immigration. Questions on birthplace were asked in the second

3The Thai Ministry of Labor kindly provided us with unit record data for these migrants, giving age, sex,

nationality, and district of registration.4The 2000 Thai Population Census identified 70,173 usual residents aged 15 and over who were born in

Cambodia, Laos, or Myanmar, and did not have Thai citizenship (our calculations from the 20 percent

sample, using sample weights.) This number should, in principle, have included irregular migrants, since

the census frame makes no reference to the legal status of residents. However, the following year, 568,245migrants aged 15 and over from these three countries registered to work in Thailand, despite a fee

equivalent to 1-2 months wages (Huguet and Punpuing 2005: 34). The Thai National Statistical Office is

aware of the poor coverage of migrants in the 2000 Census and, with funding from the World Bank, isdeveloping new procedures to improve coverage in the 2010 Census.5

Here and throughout the paper we include employees of state-owned enterprises with government

employees.


5/20

round of the survey. We exclude the 0.8 percent of respondents from this round who givea birthplace other than Thailand. It is not possible to identify non-Thais in the other three

rounds, but the number is unlikely to be high enough to have a material effect on our

results. The most important limitation of the Labor Force Survey data is that employers,the self-employed, unpaid family workers, and members of cooperatives are not asked tostate a wage. Private employees, who do provide wage data, make up 39.6 percent of the

(weighted) total. We return to this issue below. Table 1 provides summary statistics for

the whole sample and for private employees.

Initial model

Our initial model is

( )1dddd ebmw +++=

wheredw is mean log wages

6

of Thai workers in district din 2004,dm is migrant

intensity, and db is a vector of dummy variables indicating whether district dis on the

Cambodian, Lao, or Myanmar borders. Migrant intensity is defined as

( )( )ddd TMM +/log where dM is the number of registered migrants, and dT the numberof Thais, aged 15 to 59 in district din 2004. All districts are defined by 1990 rather than

2004 borders, because data for some of the control variables in subsequent specifications

are only available for 1990 borders. Our definition of migrant intensity implies that log

wages are proportional to the log of the migrant share. Most previous studies have used

the migrant share itself, rather than the log share. Our decision to use the log share ismotivated by the linear bivariate relationship between log shares and wages (Figure 2),

and the near-linear relationship between log shares and distance to the border (Figure 1).

If *dm is true migrant intensity and rd is the proportion of migrants who register, then

ddd rmm +=* . The proportion registering is an omitted variable in Equation 1 and

subsequent models. The border dummies in Equation 1 attempt to capture some of the

variation in proportions registering. Border areas contain disproportionately high

numbers of newly-arrived migrants, and migrants who commute for work who are lesslikely to be captured in registration data.

The median number of observations per district for native wages is 95.5, though 5 percentof districts have 11 observations or fewer. Estimates of the district-level wage in districtswith small numbers of observations are subject to substantial sampling error. However,

these errors should not bias our coefficient estimates since wages are an outcome

variable, rather than an explanatory variable. The variability in numbers of observations

can, however, be expected to lead to heteroskedasticiy in the error term, so all standarderrors and statistical tests presented in the paper are heteroskedasticity-robust.

6Hourly wages, including payment in kind, overtime, and bonuses.


6/20

Equation 1 and all subsequent models make no reference to the length of time thatmigrants have been in Thailand. There are no reliable longitudinal data on migrants in

Thailand that could be used to model labor market adjustment over time. Many local

labor markets had presumably partly adjusted to the presence of migrants by 2004. Ourcoefficient estimates therefore, unavoidably, reflect an unknown mix of short-run and

long-run effects.

Human capital of Thai workers

District-level differences in the mean wages of Thai workers reflect, among other things,differences in the human capital of these workers. Following Card (2005), we purge

mean wages of the effects of differences in human capital by replacing dw with a

regression-adjusted value dw . Let idw be the log wage, and idZ a vector of human capital

variables7, for person i in district d, and letd

c be a district-level fixed effect. The revised

version of Equation 1 is

( )2dddd ebmw +++=

where dw is the fitted values for dc in

( )3iddidzid ecZw +++=

EndogeneityIf the decision to come to Thailand is motivated, at least in part, by the prospect ofearning higher wages, then migrants decisions of where to live within Thailand should

also reflect wage differentials. Migrant intensity mdin Equation 2 should therefore be

treated as endogenous.

A first step towards allowing for endogenous migration is to add a vector of variables Xd

that attempts to capture determinants of wage levels.

( )4ddddd eXbmw ++++=

The first variable inXdis the distance, in hundreds of kilometers, from the center ofdistrict dto Bangkok, to capture the concentration of economic activity on the capital

city. The next variable is (log) Gross Provincial Product (GPP) per capita in the province

where district dis located8. The data refer to 1990, before large-scale migration toThailand had begun, to avoid biases that would arise if migration were influencing GPP.

7The vector consists of age, age squared, years of schooling, years of schooling squared, gender, and a full

set of second-order interactions between these terms.8

The GPP data were obtained from the website of the National Social and Economic Development Board.


7/20

Next come the proportion of the population living in urban areas, and the proportion of

households that belong to the bottom 40 percent of the household wealth d istribution.

Both measures were calculated by us from the 20 percent sample of the 1990 PopulationCensus. Household wealth was calculated by applying a principal components analysis tohousehold asset data (Filmer and Pritchett 2001). A further set of variables calculated

from the 1990 Census give the proportional distribution of the employed population by

industry, where industry is defined according to the one-digit codes for the 1958

International Classification of Industries. Finally, we include a dummy variable taking a

value of 1 if district dis in the South of Thailand (as defined by the National StatisticalOffice.) As is well-known to Thai economists, labor market outcomes differ

systematically between the South and the rest of Thailand: for instance, wages are

unexpectedly high. We do not attempt to explain these differences here.

Our next step in addressing endogenous migration is to instrument migrant intensity ondistance to the border. Let cdm be country-specific migrant intensity, defined in the

same way as cdm , except that only migrants from country c appear in the numerator.

Figure 1 shows Myanmar-specific migrant intensity versus distance to the Myanmar

border. There is a strong relationship, with a hint of curvature. In a model predicting

Myanmar-specific migrant intensity, containing border dummies andXd, distance anddistance-squared have an F-statistic of 170 (results not shown.) An equivalent regression

for Cambodian migrants gives similar results, while a regression for Laos gives a weaker

relationship. In all three cases, migrant intensity has a strong positive relationship withGPP per capita, and a strong negative relationship with the percent of households that are

poor.

[Figure 1 here]

The negative relationship between distance to the border and migrant intensity owes

something to transport costs. Migrants sometimes must pay the equivalent of severalmonths wages to be smuggled into Central Thailand (Caoutte, Archavanitkul, and Pyne

2000: 71-3). However, the particular form of the relationship also suggests that a

diffusion process may be operating. An approximately linear negative relationshipbetween migrant intensity and distance implies that migrant numbers decline

exponentially with distance. This is what would be expected if, because of friends and

neighbors effects, new migrants tend to move to districts that already contain migrants,or that are adjacent to districts already containing migrants9. The somewhat weaker

relationship between distance and migrant intensity for Lao migrants may reflect the fact

that Lao language and culture are close to those of Thais, which means that Lao migrants

can travel more easily than other migrants and rely less on migrant networks.

9A simple way to derive an exponential decline from a friends and neighbors effect is to treat districts as

equally spaced points along a line, and let migration evolve according to the following rules. In time 0,

there are no migrants. In time 1, one migrant moves to district 1. In time t> 1, the number of migrants indistrict d grows by a factor of 1+r if district d previously contained at least one migrant, grows to 1 if

district d previously contained no migrants but was next to a district with a migrant, and remains at 0

otherwise. Under these assumptions, the number of migrants declines exponentially with distance.


8/20

Whatever the exact processes that account for the association between migrant intensityand distance, it is plausible that these processes generate variation in migrant intensity

that, after controlling for distance to Bangkok, GPP per capita, household poverty, and

employment structure, is only weakly correlated with demand for labor. Instrumenting ondistance to the border should therefore help reduce biases due to the endogeneity ofmigration.

Implementation of the instrumental variables model is complicated by the use of logs in

the definition of migrant intensity. Each of the country-specific migrant intensities is well

modeled by a linear function of variables such distance to the countrys border. However,the relationship between overall migrant intensity and the three country-specific

migration intensities takes the highly non-linear form of ( )

=

c

cdd mm explog . Rather

than proceed with a complicated non-linear model, we construct instrumental variableestimates only for migrants from Myanmar. As noted above, migrants from Myanmar

account for 74 percent of all registered migrants.

Any labor market impacts from immigration to Thailand are likely to be largest for low-

skilled Thai workers, since these workers compete most directly with immigrants. We

therefore construct a low-skill version of the adjusted-wages variable, in whichindividuals are only included in Equation 3 if they have six or fewer years of schooling.

Similarly, we test for gender differences using adjusted-wage variables calculated from

males only and females only.

Results

Main results

Table 2 presents results for wages of private employees. As can be seen in column 1, andalso in Figure 2, the raw district-level relationship between migrant intensity and wages isstrongly positive. Regression-adjusting for differences in the human capital of Thai

workers in column 2 reduces the strength of the relationship slightly. Adding variables to

control for labor demand in column 3 reduces the strength considerably, though the

relationship remains positive. Column 4 is identical to column 3, except that overallmigrant intensity is replaced by Myanmar-specific migrant intensity. The number of

observations drops by 42, because of districts that do not have any migrants fromMyanmar, and which therefore do not have a defined value for Myanmar migrant

intensity. All four OLS models have negative coefficient estimates for the Myanmar

border dummy, though only the first two have negative estimates for the Lao orCambodian border.

[Figure 2 and Table 2 here]

Moving to an instrumental variables specification in column 5 reverses the sign on

migrant intensity. Because migrant intensity is defined as a log share, there is a log-log


9/20

relationship between wages and migrant numbers. The coefficient on migrant intensity incolumn 5 implies that a 10 percent increase in migrant numbers is associated with a 0.185

percent reduction in local wages. Contrary to expectations, the point estimate for low-

skilled Thai workers in column 6 is virtually identical to the estimate for all workers incolumn 5. In both cases, the Myanmar border dummy becomes statisticallyindistinguishable from zero: the overall negative relationship between migration and

wages is sufficient to explain the low wages along the Myanmar border.

Using the same approach to model wages for males and females gives an estimated

coefficient of -0.0189 (SE 0.0093) for males and -0.0143 (SE 0.0123) for females.

Extensions and robustness tests

Migration into one labor market can have spillover effects on adjacent labor markets

through, for instance, trade or migration of native workers (Borjas 2003). Moreover, evenin the absence of spillovers, a mismatch between the boundaries of districts and the

boundaries of labor markets can induce spatial correlations between districts (Anselin and

Bera 1998: 239). We investigate spatial dependency by estimating a model with spatiallags,

( )5dddddd ewWXbmw +++++=

where w is a vector containing all values ofwd, Wdis a vector of weights, and is the

spatial correlation coefficient. Wd is constructed by setting the i-th element equal to 1 if

district i shares a border with district dand 0 otherwise, and then normalizing so that Wdsums to 1. The spatial correlation coefficient governs the rate at which correlations dieoff with distance.

The spatial lags model is not designed for situations where explanatory variables are

endogenous (Kelejian and Prucha 1998: 101). We therefore replace migrant intensitywith distance to the Myanmar border. A positive coefficient on distance would imply a

negative relationship between migration and wages, though without providing

information about the magnitude of the relationship. The presence ofw on the right hand

side means that Equation 5 cannot be estimated using ordinary least squares. Instead weuse instrument on spatial lags ofXd, which allows heteroskedasticity-robust standard

errors to be calculated (Anselin and Bera 1998: 258-60)10

. We use as dependent variablesthe wages of all private employees and the wages of low-skilled private employees.

An alternative way to incorporate spillovers is to use larger geographical units (Borjas

2003). We re-estimate models 5 and 6 from Table 2 using provinces rather than districts.

[Table 3 here]

10Estimation was carried out using thestsls function in the packagespdep for the statistical languageR

(Bivand 2008).


10/20

The spatial lags models, shown in columns 1 and 2 of Table 3, fail to detect anyrelationship between wages and distance to the Myanmar border. Most of the coefficient

estimates from the spatial lags models are smaller in absolute value than those of Table 2,

but detecting a relationship with distance to the border is perhaps especially difficultbecause distance is highly correlated among neighboring districts. Without the negativerelationship between wages and migration or distance to Myanmar, the coefficients on

the Myanmar border dummy again become negative.

The instrumental variables models based on provinces, shown in columns 3 and 4, give a

completely different result. The coefficients on migrant intensity are about three times aslarge as those from Table 2, albeit with much larger standard errors.

Next we examine the effect of immigration on labor supply, measured by (i) hours

worked during the previous week by private employees, and (ii) the proportion of the

labor force (other than government employees) who were employed during the previousweek. Both supply measures are regression-adjusted for human capital and then regressed

on the same explanatory variables as column 5 of Table 2, again using distance anddistance squared to the Myanmar border as instruments for migrant intensity. Table 4

shows the results. To save space, coefficient estimates for control variables have been

omitted from the table. As can be seen in column 1, Thai employment rates appear to bepositively related to migrant intensity. Given the semi-log specification of column 1, the

coefficient on migrant intensity implies that a 10 percent increase in migrant numbers is

associated with a 0.055 percentage point increase in Thai employment rate. There is ahint of a positive relationship between migrant intensity and hours worked, but it is far

from being statistically significant.

[Table 4 here]

The wage results shown in Table 2 refer only to private employees. This raises thequestion of how immigration is affecting other workers, such as those in the informal

sector. An indirect way of assessing the effects on other workers is to examine the

relationship between migration and the sectoral distribution of Thai workers. If migrationhas a different effect on private employees than on other types of workers, then migration

should induce Thais to leave or enter private employment. We test for this possibility by

regressing the percentage of the labor force (other than government employees) working

as private employees against the same set of variables as earlier, again instrumenting on

distance to the border As can be seen in column 3 of Table 4, migration appears to havea negative relationship with private employment. The estimated coefficient and the semi-

log form of the specification imply that a ten percent increase in migrant numbers is

associated with a 0.1 percentage point reduction in the proportion of the district labor

force working as a private employee.


11/20

All previous models have excluded government employees on the grounds that migrantsdo not compete with government employees. As a robustness test, re-estimate our wage

model using (regression-adjusted) government wages as the dependent variable. A non-

zero coefficient on migrant intensity would suggest that our estimation strategy is flawed(Angrist and Krueger 2001). The estimated coefficient, shown in column 4, turns out tobe indistinguishable from zero.

Finally, we apply the same approach to testing whether immigration has induced

compensatory migration by Thais. Unlike the labor force variables, our data for the

migration variables come from the 20 percent sample of the 1990 and 2000 censuses,which provide sufficiently large samples to allow migration variables to be constructed at

the district level11.We calculate the percentage of each districts Thai population (other

than government employees) who stated that they had migrated into the district for work

within the previous two years. The mean value for 1990 is 0.85 percent and for 2000 is

1.00 percent. We also calculate a second version of this variable that includes only Thaimigrants with 6 years of schooling or less. We use district-level differences between 1990

and 2000 as our outcome variable; differencing should eliminate biases due to fixedcharacteristics that make some districts less attractive or more attractive to Thai migrants.

We assume that immigrant flows during the 1990s followed a similar geographical

pattern to that of 2004 and treat a positive coefficient on distance to the Myanmar border,or a negative coefficient on a border dummy, as evidence that Thais have avoided

migrating into districts receiving large numbers of immigrants. The results are shown in

columns 5 and 6 of Table 4. Neither column provides support for the idea thatimmigration has affected internal migration by Thais.

Discussion

The 83 percent of Thais who believe that immigration reduces local wages are, according

to our results, correct. Our preferred instrumental-variables specification, shown incolumn 5 of Table 2, has a coefficient on migrant intensity of -0.0185. Contrary to

expectations, restricting the analysis to low-skilled workers gives essentially the same

result. Immigration seems, if anything, to have slightly raised employment rates, and tohave had no effect on hours worked.

There are a number of limitations to our analysis. Immigrants are clearly attracted todistricts with higher wages, which can bias our estimate towards zero. Instrumenting on

distance to the Myanmar border should have reduced this bias, but may not have removed

it completely. One interpretation of the apparent positive relationship betweenimmigration and employment rates is that we have not completely eliminated the effects

of endogeneity..

11Also, in exploratory work, we found apparent inconsistencies in migration data from the Labor Force

Survey.


12/20

A potential weakness of analyses based on geographical variation is that they fail to takeaccount of spillovers to areas with low numbers of migrants. Our analyses incorporating

spillovers give mixed results. Adding spatial lags eliminates the relationship between

distance to the Myanmar border and wages, though this is perhaps an unreasonablydifficult test, and, in any cases, the spatial lags model implies that wages areunexpectedly low along the Myanmar border. Estimating the basic model using provinces

instead of districts has the opposite effect: it raises the point estimate for migrant

intensity, though the new estimate is imprecise.

If Thais were avoiding migrating into areas with concentrations of migrants, this wouldmask the labor market effects from immigration by diffusing the labor supply shock.

However, we find no evidence that Thai internal migration has in fact responded to

immigration. At the same time, Thais do appear to have responded to immigration by

slightly reducing their participation in private employment. This suggests that immigrants

compete most directly with private employees. It also suggests that movement betweensectors may have played a role analogous to migration in transmitting labor supply

shocks more widely across the economy.

An important limitation of our data is the extensive, but poorly understood, under-

reporting of migrant. If, despite controls for things such as distance to Bangkok andemployment structure, and border dummies, registration rates are still systematically

related to distance to the Myanmar border, then our estimates will be biased.

In summary, it would be unwise to place too much weight on the precise figure of -

0.0185. The hint of endogeneity and the large coefficient in the provinces model both

suggest that the true effect may be larger, at least for private employees. However, it isstill useful to compare this figure with estimates obtained from developed countries. Most

such estimates have come from regressing log wages on migrant share. Our estimate can

be made approximately comparable to these by dividing by the mean migrant share(Longhi et al 2005). The mean share of Myanmar migrants is 0.016, so our estimate

translates to about -1.2 from previous studies. This is a considerably larger in absolute

value than Longhi et als (2005) median value of -0.119, though smaller than the valuesof -3 to -4 found by Borjas (2003).

Even so, the impact of immigration on overall labor market outcomes in Thailand should

not be overstated. A coefficient on migrant intensity of -0.0185 implies that immigration

sufficient to double migrant numbers in 10 years would reduce wage growth by one tenthof a percentage point per year.

.


13/20

References

Altonji, J. and D. Card. 1991. The effects of immigration on the labor market outcomesof less-skilled natives. J. M. Abowd and R. B. Freeman (eds)Immigration, Trade

and Labor. Chicago: University of Chicago Press.

Angrist, J. and A. Krueger. 2001. Instrumental variables and the search for

identification: From supply and demand to natural experiments.Journal of

Economic Perspectives 15(4): 69-85.

Angrist, J. and A. Kugler. 2003. Protective or counter-productive? Labour market

institutions and the effect of immigration on EU natives. The Economic Journal

113(June): F302-F331.

Anselin, L. and A. K. Bera. 1998. Spatial dependence in linear regression models withan introduction to spatial econometrics. A. Ullah and D. Giles (eds.) Handbook of

Economic Statistics. New York: Marcel Dekker.

Bivand, R. 2008.spdep: Spatial dependence: weighting schemes, statistics, and models.

R package version 0.4-20.

Borjas, G. 2003. The labor demand curve is downward sloping: Reexamining the impact

of migration on the labor market. Quarterly Journal of Economics 118(4): 1335-1374

Caoutte, T, K. Archavanitkul, H. H. Pyne. 2000. Sexuality, Reproductive Health, and

Violence: Experiences of Migrants from Burma in Thailand. Nakhon Pathom:Institute for Population and Social Research.

Card, D. 2005. Is the New Immigration Really so Bad? Economic Journal. 115(507):

F300-F323.

Card, D, and DiNardo, J. 2000. Do immigrant inflows lead to native outflows?American Economic Review. 90(2): 36067.

Card, D., E. Lewis. 2005. The Diffusion of Mexican Immigrants During the 1990s:

Explanations and Impacts. NBER Working Paper No. 11552. Cambridge: NationalBureau for Economic Research.

Falter, J-M. No date. Minimum wages and the labor market: The case of Thailand. Paper

prepared for Country Development Partnership Social Protection, Ministry of

Labour (Thailand) and World Bank.

Filmer, D. and Pritchett, L. 2001. Estimating wealth effects without expenditure dataor

tears: An application to educational enrollments in states of India. Demography.

38(1): 115-32.

Friedberg, J. 2001. The impact of mass migration on the Israeli labor market. The

Quarterly Journal of Economics. 116(4): 1373-1408.


14/20

Friedberg, J., J. Hunt. 1996. The impact of immigrants on host country wages,employment, and growth.Journal of Economic Perspectives 9(2): 23-44

Hatton, T., M. Tani. 2005. Immigration and Inter-Regional Mobility in the UK, 1982-2000. The Economic Journal. 115(507): F342-F358.

Huguet, J., S. Punpuing. 2005.International Migration in Thailand. Bangkok:

International Organization for Migration.

Kelejian, H. and I. Prucha. 1998. A generalized spatial two-stage least squares procedure

for estimating a spatial autoregressive model with autoregressive disturbances.Journal of Real Estate Finance and Economics. 17(1): 99-121.

Longhi, S., P. Nijkamp, J. Poot. 2005. A Meta-Analytic Assessment of the Effect of

Immigration on Wages.Journal of Economic Surveys 19(3): 451477.

Mansoor A. and B. Quillin (eds.). 2007. Migrant and Remittances: Eastern Europe and

the Former Soviet Union. Washington: The International Bank for Reconstruction

and Development/ The World Bank.

Passel, J. S. 2005. Estimates of the Size and Characteristics of the Undocumented

Population. Washington: Pew Hispanic Center.

Pearson, E, S. Punpuing, A. Jampaklay, S. Kittisuksathit, A. Prohmmo. 2006. The

Mekong Challenge. Underpaid, Overworked, and Overlooked: The Realities ofYoung Migrant Workers in Thailand, Volume 1. Bangkok: International Labour

Office.


15/20

Table 1 Descriptive statistics for Labor Force Survey sample (percent)

Private employees Whole sample

Unweighted Weighted Unweighted Weighted

Female 45.9 43.2 53.3 49.9

Age 15-34 55.0 61.0 44.2 52.7

Age 35-59 45.0 39.0 55.8 47.3

6 years school or less 55.2 53.4 54.4 55.1

Private employee 100.0 100.0 35.7 39.6

Employer 0.0 0.0 3.5 3.0

Self-employed 0.0 0.0 29.0 28.2

Unpaid family worker 0.0 0.0 19.1 20.6

Government employee1 0.0 0.0 12.7 8.5

N 145,959 13,010,267 527,705 42,914,9611Includes employees of state-owned enterprises.


16/20

Table 2 The relationship between migration and wages of private employees

(1)

Raw wage,all

(OLS)

(2)

Adj wage,all

(OLS)

(3)

Adj wage,all

(OLS)

(4)

Adj wage,all

(OLS)

(5)

Adj wage,all

(IV)

(6)

Adj wage,low skill

(IV)

Migrant intensity 0.1450** 0.1166** 0.0224**

(0.0073) (0.0056) (0.0065)

Myanmar migrant intensity 0.0115* -0.0185* -0.0188*

(0.0049) (0.0092) (0.0096)On Myanmar border -0.6140** -0.4499** -0.1532** -0.1397** -0.0439 -0.0855

(0.0675) (0.0492) (0.0397) (0.0399) (0.0482) (0.0519)

On Cambodian, Lao border -0.3111** -0.2496** -0.0167 0.0296 0.0182 -0.0196

(0.0409) (0.0314) (0.0327) (0.0336) (0.0346) (0.0373)Distance to BKK, 100km -0.0447** -0.0458** -0.0511** -0.0462**

(0.0055) (0.0056) (0.0063) (0.0064)

(Log) GPP per capita 0.1240** 0.1340** 0.1863** 0.2087**(0.0196) (0.0184) (0.0203) (0.0213)

Proportion urban 0.0100 0.0057 -0.0137 -0.027

(0.0222) (0.0225) (0.0248) (0.0257)

Proportion poor -0.3330** -0.3483** -0.4282** -0.4448**(0.0663) (0.0650) (0.0685) (0.0718)

Employed in commerce -0.3238* -0.3458* -0.2242 0.1082(0.1468) (0.1435) (0.1677) (0.2214)

Employed in construction 1.0917* 1.1192* 1.5110** 0.9095

(0.4649) (0.4674) (0.4760) (0.4685)Employed in electric -0.1436 -0.2879 -0.5043 -0.7988*

(0.2910) (0.2805) (0.3160) (0.3841)

Employed in manufacturing 0.3081** 0.2994** 0.2979** 0.3237**

(0.0955) (0.0941) (0.0975) (0.1024)Employed in mining -1.8911** -1.8028** -0.9679 -1.5434*(0.6355) (0.5856) (0.6229) (0.7187)

Employed in services 0.4012** 0.4397** 0.5119** 0.4322**

(0.1241) (0.1204) (0.1311) (0.1368)Employed in transport -0.9942 -1.0575 -2.2586** -1.7678*

(0.6008) (0.6042) (0.6691) (0.7923)

Employed in other 2.6139** 2.7819** 2.6789** 1.4547

(0.8599) (0.8381) (0.8745) (0.9029)

South 0.4040** 0.4058** 0.4689** 0.5318**(0.0312) (0.0328) (0.0391) (0.0399)

(Constant) 3.7834** 0.1341** -0.6387** -0.7143** -1.0370** -1.1816**

(0.0434) (0.0315) (0.0917) (0.0822) (0.1084) (0.1124)

Adjusted R-squared 0.4211 0.4325 0.6713 0.6838 0.6697 0.6915

N 760 760 760 718 718 717

Note the omitted category for the employment variables is employment in agriculture.

*Significant at 5% level. **Significant at 1% level. All standard errors are heteroskedasticity-robust.


17/20

Table 3 Models of wages allowing for spatial dependency

(1)

All, districts

(spatial lag)

(2)

Low-skill,districts

(spatial lag)

(3)

All, provinces

(IV)

(4)

Low-skill,

provinces (IV)

Myanmar migrant intensity -0.0566* -0.0504*

(0.0251) (0.0255)

Distance to Myanmar border (100km) 0.0050 0.0003

(0.0060) (0.0063)On Myanmar border -0.0505 -0.0961** 0.0599 0.0175

(0.0336) (0.0361) (0.0653) (0.0660)

On Cambodian, Lao border 0.0014 -0.0076 0.0018 0.0057

(0.0302) (0.0309) (0.0371) (0.0368)Distance to Bangkok (100km) -0.0284** -0.0283** -0.0596** -0.0568**

(0.0068) (0.0070) (0.0129) (0.0125)

(Log) GPP per capita 0.0869** 0.1051** 0.2235** 0.2046**(0.0199) (0.0222) (0.0489) (0.0503)

Proportion urban 0.0337 0.0202 0.0996 0.048

(0.0213) (0.0222) (0.0732) (0.0761)

Proportion poor -0.3683** -0.3580** -0.5049* -0.6083**(0.0604) (0.0624) (0.2209) (0.2186)

Employed in commerce -0.2970 -0.0462 1.1387 1.4833(0.1517) (0.2031) (0.9301) (0.9555)

Employed in construction 0.9331* 0.7764* 4.2784** 3.0102*

(0.3971) (0.3798) (1.4290) (1.4063)Employed in electric -0.1391 -0.5365 0.5237 1.2822

(0.2768) (0.3202) (4.0947) (3.7929)

Employed in manufacturing 0.1816* 0.1813* 0.8493** 0.9364**

(0.0786) (0.0866) (0.3054) (0.3151)Employed in mining -0.749 -1.4417** -7.5262* -7.5629*(0.5142) (0.5412) (3.2895) (3.4983)

Employed in services 0.2877** 0.2206 1.4192** 1.2271

(0.1060) (0.1145) (0.5405) (0.6478)Employed in transport -0.9328 -0.6162 -17.3680** -14.6603*

(0.5047) (0.5931) (5.4096) (6.0873)

Employed in other 1.0470 0.5168 5.0849 2.4833

(0.7734) (0.7883) (2.6400) (2.7469)

South 0.2832** 0.3448** 0.5122** 0.5910**(0.0415) (0.0496) (0.0852) (0.0840)

(Constant) -0.4624** -0.5792** -1.4512** -1.3651**

(0.0947) (0.1082) (0.3305) (0.3446)

Spatial correlation coefficient 0.3537** 0.3259**

(0.0760) (0.0812)

Adjusted R-squared 0.8303 0.8456

N 767 767 73 73

Note the omitted category for the employment variables is employment in agriculture.



18/20

Table 4 Relationship between migration and labor market outcomes and migration

(1)

Percent

employed

(IV)

(2)

Hours last

week

(IV)

(3)

Percentprivate

employees

(IV)

(4)

Government

wages

(IV)

(5)

Change in %

migrants, all

(OLS)

(6)

Change in %migrants,

low-skill

(OLS)

0.5527** 0.3378 -0.9675* -0.0060Myanmar migrant

intensity (0.1574) (0.1972) (0.4509) (0.0078)

0.0093 0.0064Distance to Myanmar

border (100km) (0.0264) On Myanmar border -0.8458 1.5468 -0.0578 -0.0576 -0.116 -0.0942

(0.7194) (0.8968) (2.4055) (0.0370) (0.1875) (0.1326)

0.4958 0.5145 -2.4009 -0.0023 0.1404 0.0052On Cambodian, Lao

border (0.4011) (0.8842) (1.3245) (0.0263) (0.0960) (0.0654)Note these coefficients estimates were obtained from models with the same control variables as Table 2.



19/20

Figure 1 - Myanmar-specific migrant intensity versus distance to the Myanmar

border


20/20

Figure 2 Wages versus migrant intensity

Documents

Does immigration to Thailand reduce the wages of Thai workers?